- Email: [email protected]
WO3–SiO2 for selective catalytic reduction of NO by CO
Applied Catalysis B: Environmental 84 (2008) 420–425
Contents lists available at ScienceDirect
Applied Catalysis B: Environmental journal homepage: www.elsevier.com/locate/apcatb
Formation of active sites on Ir/WO3–SiO2 for selective catalytic reduction of NO by CO Tetsuya Nanba *, Satoru Shinohara, Shouichi Masukawa, Junko Uchisawa, Akihiko Ohi, Akira Obuchi National Institute of Advanced Industrial Science and Technology (AIST), Research Center for New Fuels and Vehicle Technology, 16-1 Onogawa, Tsukuba 305-8569, Japan
A R T I C L E I N F O
A B S T R A C T
Article history: Received 10 January 2008 Received in revised form 18 April 2008 Accepted 22 April 2008 Available online 8 May 2008
Pretreatment conditions for the activation of Ir/WO3–SiO2 for the selective catalytic reduction of NO by CO in the presence of excess O2 were studied. Sequential treatment involving calcination in the presence of O2 and H2O followed by reduction and then re-oxidation under mild conditions was found to effectively activate Ir/WO3–SiO2. Temperature-programmed desorption during calcination, X-ray diffraction, and temperature-programmed reduction by H2 revealed that calcination was necessary for oxidative removal of the NH3 ligands from the iridium precursor, that reduction produced metallic iridium and partially reduced tungsten oxide, and that re-oxidation produced tungsten oxide with low reducibility. Transmission electron microscopy revealed that Ir was supported on finely dispersed tungsten oxide and that the iridium particle size after the sequential activation was 1–1.5 nm. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Nitrogen oxide Carbon monoxide Iridium Selective catalytic reduction Tungsten oxide Silica
1. Introduction Selective catalytic reduction (SCR) of NOx (NO + NO2) by reductants present in exhaust gases is an attractive way to control NOx emissions from diesel and lean-burn engines. Since Iwamoto et al. [1] and Held et al. [2] discovered SCR by hydrocarbons (HCSCR), many papers on HC-SCR have been published [3]. Many kinds of catalysts, such as platinum-group metals [4,5], zeolites [6,7], and Al2O3-supported transition metals [8,9], exhibit HC-SCR activity. However, no HC-SCR catalytic converter system is yet in practical use. A new type of internal combustion engine is being developed for the reduction of NOx emissions: the homogeneous charge compression ignition (HCCI) engine. This engine shows low NOx and particulate emissions and relatively high-CO emissions [10]. Therefore, the use of CO as a reductant for exhausts with low-NOx concentrations is of interest. Although CO acts as a reductant in the three-way catalyst [11], it is a poor reductant under lean conditions [12,13]. Recently, iridium catalysts have been reported to show CO-SCR activity. For example, Ogura et al. reported that Ir/ silicalite at a very low-Ir loading exhibits high activity in the absence and presence of SO2 [14]. Haneda et al. reported that Ir/ SiO2 shows high activity only in the presence of SO2 [15].
* Corresponding author. Tel.: +81 29 861 8288; fax: +81 29 861 8259. E-mail address: [email protected] (T. Nanba). 0926-3373/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2008.04.027
Shimokawabe et al. reported that Ir/WO3 and Ir/ZnO show high activities [16] and that Ir/WO3 is effective even in the presence of SO2 [17]. Because the sulfur content in diesel fuel is now severely restricted in many countries, a catalyst having CO-SCR activity in the absence of SO2 is desirable. We have recently found that the combination of WO3 and SiO2 is an effective catalyst support [18]. The Ir/WO3–SiO2 is composed of Ir/WO3 highly dispersed on SiO2. In this study, we attempted to optimize the preparation conditions for the formation of active sites during preparation of the supported catalyst, and we discuss the active state of the Ir/WO3. 2. Experimental 2.1. Catalyst preparation The combined oxide support, WO3–SiO2, was prepared by two methods. Method 1. Addition of metallic tungsten powder (Mitsuwa Chemical Co.) to 15% H2O2 yielded a peroxopolytungstic acid solution [19]. After the pH of this solution was adjusted to approximately 8.5 with NH3, a predetermined amount of silica sol was added (Cataloid S-20L, containing 20 wt% SiO2; Catalysis & Chemical IND Co.). The mixture was dried at 110 8C and calcined in air at 500 8C for 4 h. Method 2. Ammonium tungstate parapentahydrate (Wako Pure Chemical Industries) [(NH4)10W12O41 5H2O], was dissolved in 0.7 M malic acid aqueous solution. A predetermined amount of silica sol was added, and the pH was adjusted to approximately 8.8 with NH3. Gelation occurred within
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
421
1 h after the addition of NH3. The gel was dried at 100 8C and calcined in air at 500 8C for 4 h. The WO3–SiO2 made by method 1 was prepared with WO3 contents ranging from 0% to 100%. The mixed oxide support is designated WO3(x)–SiO2-1 or -2, where the value in parentheses is the weight percentage of WO3 in the support and the next number is the preparation method. Iridium supported on WO3–SiO2 was prepared by the impregnation method using Ir(NH3)6(OH)3 as a precursor. The loading of iridium was adjusted from 0.25 to 10 wt%. The impregnated supports were calcined at 500 8C for 4 h under various atmospheres.
flow of 2% H2/Ar at a heating rate of 10 8C/min after oxidation of the sample in a 5% O2 flow at 500 8C for 2 h. The morphologies of Ir and WO3 particles were observed by transmission electron microscopy (TEM; JEM2000EX; JEOL). Temperature-programmed desorption (TPD) of nitrogen-containing compounds was carried out to clarify how the Ir precursor decomposed during these calcination procedures. After Ir loading and drying, 0.1 g of the sample was placed in the apparatus used for the activity tests. The TPD profiles were obtained from room temperature to 500 8C in a 100 mL/min flow of 10% H2, 20% O2, or 20% O2 + 1% H2O at a heating rate of 5 8C/min.
2.2. Activity tests
3. Results
Catalytic activity tests were conducted in a conventional fixedbed reactor at atmospheric pressure. The catalyst (0.1 g; sieved to between 0.15 and 0.25 mm) was diluted with quartz sand or SiC granules (particle size, 0.3–0.5 mm). The flow rate of the feed gas was 225 or 400 mL/min, corresponding to a space velocity of 34,000 or 60,000 h1, respectively. The feed gas composition for the 225 mL/min flow rate was 500 ppm NO, 5000 ppm CO, 10% O2, and 1% H2O, with He as a balance gas; for the 400 mL/min flow rate, the feed gas composition was 500 ppm NO, 3000 or 5000 ppm CO, 10% O2, and 1% or 6% H2O, with N2 as a balance gas. Before most of the catalytic activity tests, the catalysts were reduced in 10% H2 at 600 8C for 1 h. The temperature dependence of the catalytic activity was measured from 400 to 180 8C in steps of 10 8C. At each temperature, the effluent gas was sampled after the temperature had been maintained for 1 h. In the other activity test, in which measurements were carried out at a constant temperature of 270 8C, the samples were pretreated under various conditions after the H2 reduction. Concentrations of NO, NO2, N2O, CO, and CO2 in the reactant and product gases were analyzed with an FT-IR spectrometer (Magna 560; Nicolet) equipped with a multireflectance gas cell (Gemini Specialty Optics, optical path length, 2 m) and a mercury–cadmium–telluride detector. For the feed gas balanced with He, a gas chromatograph (M200; Agilent) equipped with an MS-5A PLOT column (for N2 and CO) or a PoraPLOT Q column (for CO2 and N2O) and a thermal conductivity detector was also used. The catalytic activity was evaluated in terms of NOx conversion and N2 selectivity. To ensure accuracy, N2 selectivity was evaluated only when the NOx conversion was above 10%:
3.1. Catalytic activity
NOx conversion ¼
½inlet NOx ½outlet NOx 100 ½inlet NOx
N2 selectivity ðfor He balanceÞ ¼
½outlet N2 100 ½outlet N2 þ ½outlet N2 O
N2 selectivity ðfor N2 balanceÞ ½outlet N2 O 100 ¼ 12 ½inlet NOx ½outlet NOx
2.3. Characterization The Brunauer–Emmett–Teller (BET) specific surface area was measured by N2 adsorption under a flow condition (Model 4232; Nikkiso). X-ray diffraction (XRD) patterns were measured at 40 kV and 100 mA of X-ray power (RU-300; Rigaku). Identification of crystal phase was carried out by using the JCPDS database. Temperature-programmed reduction by H2 (H2-TPR) was carried out in an adsorption test apparatus equipped with a thermal conductivity detector (ADT700; Ohkura-Riken). The TPR profiles were obtained from room temperature to 600 8C in a 100 mL/min
Table 1 lists the CO-SCR activities of 0.5 wt% Ir/WO3–SiO2-1 with various WO3 contents. At WO3 contents above 50%, the maximum NOx conversion and N2 selectivity at the temperature of maximum NOx conversion were almost the same as those of Ir/ WO3. At WO3 contents below 30%, the activity was markedly enhanced. Note that a high activity was attained even at 1 wt% WO3. WO3 contents of 1–30 wt% correspond to W/Ir molar ratios between 1.7 and 50. Note also that the precursor of WO3 had a negligible influence on the CO-SCR activity. We confirmed that the crystal phases of the WO3 in the Ir/WO3–SiO2-1 and -2 samples were the same, monoclinic and orthorhombic. The fact that the two samples were composed of the same crystal phases indicates that the WO3 precursor probably had no influence on the crystal phase. Examination of the activities at various Ir loadings showed that the maximum NOx conversion was achieved at 0.5 wt% (Table 2), and CO conversion increased with increasing Ir loading. N2 selectivity decreased with increasing Ir loading. 3.2. Effect of calcination, reduction, and additional pretreatment conditions We investigated the effects of calcination conditions on the activity of 0.5 wt% Ir/WO3(10)–SiO2-2 in terms of the maximum Table 1 Effect of WO3 content on CO-SCR activity WO3 content (wt%)
Temperature (8C)
NOx conversion (%)
N2 selectivity (%)
0 1 10 30 40 50 80 100
250 260 260 265 265 255 275 270
11 85 86 82 69 49 42 39
81 89 89 87 81 79 72 79
Catalyst: 0.5 wt% Ir/WO3(x)–SiO2-1; catalyst weight: 0.1 g; flow rate: 225 mL/min; composition: 500 ppm NO, 5000 ppm CO, 10% O2, 1% H2O, and balance He. Table 2 Effect of iridium loading on CO-SCR activity Loading (wt%)
NOx conversion (%)a
CO conversion (%)
N2 selectivity (%)
0.25 0.5 1 2 5 10
21 34 25 22 21 18
37 51 79 91 91 95
68 68 58 51 45 41
Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-2; catalyst weight: 0.1 g; flow rate: 400 mL/ min; composition: 500 ppm NO, 3000 ppm CO, 10% O2, 6% H2O, and balance N2. a Maximum NOx conversion. CO conversion and N2 selectivity were measured at 280 8C.
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
422 Table 3 Effect of calcination condition on CO-SCR activity Calcination condition
Reaction temperature (8C)
NOx conversion (%)
CO conversion (%)
N2 selectivity (%)
Air, 500 8C 20% O2, 500 8C 20% O2 + 1% H2O, 500 8C 1% O2 + 6% H2O, 500 8C 6% H2O, 500 8C 10% H2, 600 8Ca
275 260 280 275 330 290
34 7 26 31 10 4
51 81 66 72 98 97
68 – 65 68 48 –
Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-2; catalyst weight: 0.1 g; flow rate: 400 mL/min; composition: 500 ppm NO, 3000 ppm CO, 10% O2, 6% H2O, and balance N2. After calcination, sample was reduced in H2 at 600 8C for 1 h, and then activity was measured as the manner of temperature dependence. a Reduction procedure as pretreatment was omitted.
Table 4 Effect of reduction condition on CO-SCR activity Reduction condition
10% H2, Nona 10% H2, 10% H2, 10% H2,
600 8Ca b
400 8C 500 8Cb 600 8Cb
NOx conversion (%)
CO conversion (%)
N2 selectivity (%)
81 33 26 29 31
82 66 61 71 72
88 72 62 64 68
Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-1; catalyst weight: 0.1 g; calcination condition was in air at 500 8C for 4 h. Activity was measured as the manner of temperature dependence. a Flow rate: 225 mL/min; composition: 500 ppm NO, 5000 ppm CO, 10% O2, 1% H2O, and balance He. Reaction temperature was 270 8C. b Flow rate: 400 mL/min; composition: 500 ppm NO, 3000 ppm CO, 10% O2, 6% H2O, and balance N2. Reaction temperature was 280 8C.
NOx conversion as well as the CO conversion and N2 selectivity at the temperature of maximum NOx conversion (Table 3). Among the catalysts calcined in various atmospheres, the catalysts calcined in ambient air showed the highest NOx conversion. The catalyst calcined in 20% O2 showed very low activity, whereas the catalyst calcined in 20% O2 + 1% H2O showed higher activity. Furthermore, the catalyst calcined in 1% O2 + 6% H2O showed activity comparable to that of the catalyst calcined in air, whereas the catalyst calcined in 6% H2O showed low activity. The catalyst calcined in 10% H2 also showed low activity. The catalysts showing low-NOx conversions exhibited relatively high-CO conversions. These results suggest that the co-existence of O2 and H2O was necessary for activation of the catalyst. We compared the activities of reduced and unreduced catalysts after calcination and found that the reduced catalysts showed markedly higher NOx conversion than the unreduced catalysts (Table 4). The maximum NOx conversions over the catalysts reduced at 400, 500, and 600 8C were 26%, 29%, and 31%, respectively, which suggests that the reduction temperature had little effect on the activity. We investigated the effects of additional treatment conditions on the activity of 0.5 wt% Ir/WO3(10)–SiO2-2, which had been calcined in air and then reduced at 600 8C (Table 5). The catalytic activity was evaluated at 270 8C. When the catalysts were cooled
from 400 to 270 8C under a flow of CO-SCR feed gas, the catalysts showed high-NOx conversions. In contrast, cooling the catalysts under a flow of He resulted in only 7% NOx conversion. However, the catalysts subjected to additional treatment at 400 8C in 10% O2 showed approximately 50% NOx conversion. This result suggests that re-oxidation after reduction effectively activated the catalysts. However, the catalyst pretreated at 600 8C in 100% O2 was almost inactive, suggesting that mild re-oxidation was better for activation than severe re-oxidation. 3.3. Characteristics of Ir/WO3–SiO2 The specific surface area of Ir/SiO2 was 190 m2/g, and the value decreased as the WO3 content increased (Fig. 1). For the samples with 10–80 wt% WO3, the specific surface area decreased linearly with increasing WO3. The specific surface area in the range of 80– 100 wt% WO3 also linearly decreased with increasing WO3. Note that the molar amounts of WO3 and SiO2 are almost equivalent at 80 wt% WO3. The dashed line in Fig. 1 indicates the surface area of SiO2 contained in 1 g of each Ir/WO3–SiO2 catalyst. In the range from 10 to 80 wt% WO3, the slope of the curve for the SiO2 surface area is the same as the slope of the curve for the catalysts. These results suggest that the change in specific surface area in this region depended only on the SiO2 content. We measured XRD profiles of 0.5 wt% Ir/WO3(10)–SiO2-2 after calcination in air, after reduction at 600 8C, and after the CO-SCR activity test (Fig. 2). The sample calcined in air showed peaks due to WO3 (JCPDS: 43-1035, 20-1324) (Fig. 2(a)), which suggests that WO3 over SiO2 was crystalline even at 10 wt% WO3. For the sample reduced at 600 8C, the WO3 diffraction peaks disappeared, and peaks ascribed to WO3x (0 < x < 1) (JCPDS: 30-1387), WO2 (JCPDS: 32-1393), and W metal (JCPDS: 04-0806) were observed (Fig. 2(b)). For the sample subjected to CO-SCR, the XRD profile exhibited broad peaks in the WO3 region, and the WO2 and W metal peaks diminished (Fig. 2(c)). The presence of the broad peaks in the WO3 region (broader than those in Fig. 2(a)) suggests that fine WO3 particles were formed. In contrast, the sample subjected to severe oxidation conditions (100% O2 at 600 8C) showed a sharp peak at 23.18 (Fig. 2(d)), which is attributed to WO3; this treatment did not restore the peak intensity to the original level (Fig. 2(a)).
Table 5 Effect of treatment condition after reduction on CO-SCR activity Condition CO-SCR feed (introduced at 400 8C) Non CO-SCR feed (400 8C, 1 h) 10% O2 (400 8C, 1 h) 100% O2 (600 8C, 1 h)
Atmosphere before reaction CO-SCR feed He He He He
a
NOx conversion (%)
CO conversion (%)
N2 selectivity (%)
84 7 49 54 2
84 97 73 97 3
87 – 72 74 –
Activity measurement was carried out at 270 8C. Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-2; catalyst weight: 0.1 g; flow rate: 225 mL/min; composition of CO-SCR feed gas: 500 ppm NO, 5000 ppm CO, 10% O2, 1% H2O, and balance He. Calcination: 500 8C for 4 h in air; reduction: 600 8C for 1 h in 10% H2. a Temperature dependence procedure as shown in Fig. 1(a).
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
423
Fig. 1. Dependence of BET surface area on WO3 content for 0.5 wt% Ir/WO3(x)– SiO2-1.
Although crystallization of WO3 seemed to be promoted by the severe oxidation conditions, the WO3 particles obtained after calcination, reduction, and severe re-oxidation must have been smaller than those obtained after calcination alone. We also examined the H2-TPR profiles for various 0.5 wt% Ir catalysts (Fig. 3). Ir/SiO2 exhibited a reduction peak at 225 8C, which was assigned to the reduction of IrO2 to Ir metal. Ir/WO3 had a broad peak centered at 440 8C, in addition to a peak at 220 8C, whereas WO3 had no peak below 550 8C. The reduction temperature of WO3 is known to be lowered in the presence of a noble metal [20,21]. Therefore, we ascribed the broad peak at around 440 8C to the reduction of WO3. H2 consumption values for the peaks at around 220 8C for Ir/SiO2 and Ir/WO3 were 42 and 310 mmol/g, respectively. These values correspond, respectively, to 0.8 and 6 times the calculated H2 consumption for
Fig. 2. XRD profiles of 0.5 wt% Ir/WO3(10)–SiO2-2 treated under various conditions: (a) after calcination in air, (b) after reduction at 600 8C, (c) after re-oxidation under CO-SCR conditions, and (d) after re-oxidation at 600 8C in 100% O2.
Fig. 3. H2-TPR profiles of various 0.5 wt% Ir catalysts.
stoichiometric reduction of IrO2 to Ir. The high-H2 consumption (substantially exceeding unity) suggests that IrO2 reduction over WO3 proceeded together with reduction of WO3. Ir/WO3(10)– SiO2-2 exhibited H2 reduction peaks at 225 and 350 8C. The amount of H2 consumed below 500 8C corresponded to the formation of WO2.9. Additional H2 was consumed above 550 8C, which suggests that tungsten oxides were reduced further at high temperatures. Re-oxidized Ir/WO3(10)–SiO2-2 exhibited a H2 reduction peak at 190 8C, which we ascribed to the reduction of IrO2 and WO3; no other peak appeared below 550 8C, which suggests that the reduction of tungsten oxides was suppressed after the re-oxidization. We investigated the H2-TPR profiles of Ir/WO3–SiO2 catalysts with various WO3 contents (Fig. 4(a)). At WO3 contents below 10 wt%, the TPR profiles showed a peak at 235 8C with a shoulder at 200 8C. At WO3 contents above 10 wt%, the shoulder at 200 8C increased and shifted toward lower temperatures as the WO3 content increased. We determined the molar ratio of H2 consumed to Ir atoms for the peak at around 220 8C, including the shoulder (or peak) at around 200 8C, and examined the correlation between this ratio and the WO3 content in 0.5 wt% Ir/WO3–SiO2-1 after calcination (Fig. 4(b)). A ratio of 2 suggests stoichiometric reduction of IrO2 to Ir. Below 30 wt% WO3, the ratio was slightly lower than 2 at 30 wt% WO3, the ratio was slightly higher than 2, and as the WO3 content increased, the ratio increased markedly and substantially exceeded 2. This observed behavior for the peak and the shoulder at > 30 wt% WO3 indicates that the reducibility of WO3 in the vicinity of IrO2 was higher for catalysts containing >30 wt% WO3 than for catalysts containing less WO3. Lucas et al. reported that the reducibility of WO3 decreases with decreasing WO3 content [22]. These results suggest that reduction of WO3 in the vicinity of IrO2 did not occur to a substantial extent for catalysts with WO3 contents below 30 wt% and that the reduction of WO3 increased with increasing WO3 content at values above 30 wt%.
424
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
Fig. 4. H2-TPR profiles of the Ir/WO3–SiO2-1 (a) and the molar ratio of H2 consumed to Ir atoms for the peak at 220 8C and the dependence of the ratio on WO3 content (b).
Fig. 5. TEM image of 0.5 wt% Ir/WO3(10)–SiO2-2 treated by optimal sequential treatment (a) and reduced without calcination (b).
TEM images of 0.5 wt% Ir/WO3(10)–SiO2-2 subjected to CO-SCR conditions showed that cubic tungsten oxides with particle sizes larger than 5 nm were dispersed on SiO2 and that Ir particles 1– 1.5 nm in size were located on the tungsten oxides (Fig. 5(a)). In contrast, TEM images of the reduced 0.5 wt% Ir/WO3(10)–SiO2-2 without calcination, which had low activity, exhibited WO3 particles of various sizes and Ir particles 2–5 nm in size (Fig. 5(b)). Furthermore, there were some Ir particles that were not associated with WO3 particles, as indicated by the absence of a lattice pattern of WO3 around the Ir particles. We concluded that the final active form of Ir/WO3–SiO2 obtained by sequential treatment involving calcination, reduction, and re-oxidation was very fine Ir/WO3 particles dispersed on SiO2. 4. Discussion The preferred composition of Ir/WO3–SiO2 for CO-SCR was 1– 30 wt% WO3 in the support and a 0.5 wt% Ir loading. The surface area of Ir/WO3–SiO2 was linearly correlated with SiO2 content. In contrast, almost equivalent activities of the samples with 1– 30 wt% WO3 loading were observed. Previously, we reported that iridium particles existed selectively on the WO3 particles, and we supposed that the increase in the surface area of Ir/WO3 by dispersion on the SiO2 resulted in the enhancement of the CO-SCR activity [18]. However, these results suggest that the enhancement of the CO-SCR activity of Ir/WO3–SiO2 at lower WO3 contents was due not to an increase in the Ir/WO3 surface area but to some other reason, which is discussed below.
Fig. 6. Temperature-programmed desorption of nitrogen-containing compounds during calcination under 10% H2, 20% O2, and 20% O2 + 1% H2O. Catalyst weight, 0.1 g; flow rate, 400 mL/min; ramping rate, 5 8C/min.
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
The change in CO-SCR activity with WO3 content agreed well with the change in the reduction properties of IrO2 and WO3 at 220 8C as revealed by H2-TPR (Fig. 4). This result suggests that what makes WO3 preferable as a support for Ir is its resistance to reduction when in the vicinity of Ir particles. To activate Ir/WO3–SiO2, the catalysts had to be treated sequentially by calcination (oxidation in the presence of H2O), reduction, and re-oxidation. TEM observations indicated that the reduced sample without calcination had larger Ir particles than the sample subjected to the optimal sequential treatment and that the former sample had separate Ir and WO3 particles. Ir/SiO2 is known to exhibit activity only in the presence of SO2 [15], so the decrease in the CO-SCR activity of the reduced catalysts without calcination was probably due to dissociation of Ir from WO3. To observe the decomposition behavior of Ir precursor, we used TPD to measure the formation of nitrogen-containing compounds from WO3–SiO2 impregnated with Ir(NH3)6(OH)3 (Fig. 6). NOx and N2O formed at 375 8C in 20% O2. Furthermore, the addition of H2O to 20% O2/N2 led to the evolution of NOx at a lower temperature of approximately 335 8C. This result suggests that calcination in the presence of O2 and H2O resulted in the H2O-promoted oxidative removal of ligands of Ir(NH3)6(OH)3. On the other hand, calcination under H2/ N2 resulted in NH3 formation between 200 and 300 8C, which suggests the simple release of NH3 ligands. Considering that H2 reduction after calcination was performed in the absence of NH3 ligands for the active Ir/WO3–SiO2, the residual NH3 ligands might relate to the dissociation of Ir form WO3 in the reducing conditions. The role of calcination is suspected to be the oxidative removal of NH3 ligands. H2 reduction after calcination was necessary for activating Ir/ WO3–SiO2. Metallic iridium is believed to be the active site for HCSCR [23,24] and CO-SCR [15], and the reduction procedure is necessary for the formation of metallic iridium. Moreover, the reduction procedure markedly changed the oxidation state of tungsten, as shown by the XRD results. Although WO3 is known to form alloys with various metals upon reduction [21] and an Ir–W alloy has been shown to have high resistance against oxidation [25], 600 8C is too low a temperature for the formation of an Ir–W alloy [26]. We observed only a small difference between the activities of the samples reduced at 400 8C to 600 8C, and we also confirmed that W metal and WO2 were not observed by XRD after reduction at 400 8C. This result suggests that severe reduction to form W and WO2 was not dispensable. H2 consumption below 500 8C in H2-TPR of Ir/WO3(10)–SiO2-2 was insufficient for the formation of WO2. Therefore, we suggest that the necessary oxidation state of tungsten after reduction is WO3x (0 < x < 1). We believe that the re-oxidation step was necessary for the formation of WO3 with low reducibility as a support for iridium. Although the H2-TPR profile of Ir/WO3–SiO2 after calcination showed a broad peak at 350 8C, the peak disappeared in the H2TPR profile of Ir/WO3–SiO2 after re-oxidation. H2 consumption at around 200 8C in the H2-TPR measurement of the re-oxidized sample was about 0.4 times the amount expected for stoichiometric IrO2 reduction. These results suggest that the iridium particles were composed mainly of metallic iridium and that the reducibility of WO3 was decreased by re-oxidation. Bigey et al. have reported that re-oxidized tungsten oxides exhibit lower reducibility than fresh WO3 [21]. It is known that the oxidation state of tungsten is easily changed during reactions such as cracking and isomerization of hydrocarbons and that retaining a stable oxidation state of tungsten with high activity is difficult [20]. We suspect that the decrease in the reducibility of WO3 was important for maintaining an active metallic iridium surface. It is concluded that the final active site formed by sequential treatments was metallic-state iridium finely dis-
425
persed on WO3 having small particle size and resistance to reduction. 5. Conclusions Ir/WO3–SiO2 showed high activity for selective NO reduction by CO. We attempted to optimize the conditions for preparation of the catalyst. The optimum composition was 0.5 wt% Ir and 1–30 wt% WO3 in the support. At the optimum WO3 contents, NOx conversions were almost the same for all WO3 contents. The surface area of Ir/WO3–SiO2 depended on SiO2 content. Although the combination of WO3 and SiO2 as a support was effective for dispersing Ir/WO3, the surface area did not influence the CO-SCR activity. The H2-TPR results indicated that the catalysts with the optimum WO3 content had lower reducibility of WO3 than the catalysts with a high-WO3 content. Pretreatment after Ir loading (that is, calcination, reduction, and additional treatment) strongly influenced the CO-SCR activity. The optimum sequence of pretreatment was calcination in the presence of O2 and H2O followed by H2 reduction and then reoxidation under mild oxidizing conditions. TPD of nitrogencontaining compounds during calcination suggested that the NH3 ligands in the Ir precursor were removed by oxidation in the presence of O2 but that NH3 release under a reducing condition occurred at a lower temperature than oxidative removal of the NH3 ligands. H2O addition lowered the temperature for oxidative removal of the NH3 ligands. During the reduction, iridium was reduced to the metallic state, and WO3 was also reduced to metal and WO2. The reduction of WO3 at least to WO3x (0 < x < 1) was required to activate the catalyst. H2-TPR results suggested that reoxidation suppressed the reducibility of WO3. The re-oxidation step formed the favorable state of tungsten oxide as a support for iridium. XRD and TEM results suggested that very small WO3 particles were dispersed on SiO2. We concluded that the active state of the Ir/WO3–SiO2 catalysts, which was formed by sequential treatments, was metallic-state iridium finely dispersed on WO3, which had small particle size and resistance to reduction. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]
M. Iwamoto, H. Yahiro, S. Shundo, Y. Yu-u, N. Mizuno, Appl. Catal. 69 (1991) L15. W. Held, A. Ko¨nig, T. Richter, L. Puppe, SAE Tech. Pap. Ser. No. 900496 (1990). V.I. Paˆrvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233. A. Obuchi, A. Ohi, M. Nakamura, A. Ogata, K. Mizuno, H. Ohuchi, Appl. Catal. B 2 (1993) 71. R. Burch, T.C. Watling, J. Catal. 169 (1997) 45. J.L. d’Itri, W.M.H. Sachtler, Catal. Lett. 15 (1992) 289. S. Sato, Y. Yu-u, H. Yahiro, N. Mizuno, M. Iwamoto, Appl. Catal. 70 (1991) L1. H. Hamada, Y. Kintaichi, M. Inaba, M. Tabata, T. Yoshinari, H. Tsuchida, Catal. Today 29 (1996) 53. Y. Torikai, H. Yahiro, N. Mizuno, M. Iwamoto, Catal. Lett. 9 (1991) 91. D. Jun, K. Ishi, N. Iida, JSME Int. J. Ser. B 46 (2003) 60. J.T. Kummer, J. Phys. Chem. 90 (1986) 4747. M.L. Unland, J. Phys. Chem. 77 (1973) 1952. M.L. Unland, J. Catal. 31 (1973) 459. M. Ogura, A. Kawamura, M. Matsukata, E. Kikuchi, Chem. Lett. (2000) 146. M. Haneda, Pursparatu, Y. Kintaichi, I. Nakamura, M. Sasaki, T. Fujitani, H. Hamada, J. Catal. 229 (2005) 197. M. Shimokawabe, N. Umeda, Chem. Lett. 33 (2004) 534. M. Shimokawabe, M. Niitsu, H. Inomata, N. Iwasa, M. Arai, Chem. Lett. 34 (2005) 1426. T. Nanba, S. Shinohara, J. Uchisawa, S. Masukawa, A. Ohi, A. Obuchi, Chem. Lett. 35 (2006) 450. J. Oi, A. Kishimoto, T. Kudo, J. Solid State Chem. 103 (1993) 176. C. Bigey, L. Hilaire, G. Maire, J. Catal. 198 (2001) 208. C. Bigey, L. Hilaire, G. Maire, J. Catal. 184 (1999) 406. A. de Lucas, J.L. Valverde, P. CanA¨izares, L. Rodriguez, Appl. Catal. A 172 (1998) 165. M. Nawdali, E. Iojoiu, P. Ge´lin, H. Praliaud, M. Primet, Appl. Catal. A 220 (2001) 129. C. Wo¨gerbauer, M. Maciejewski, A. Baiker, J. Catal. 205 (2002) 157. Y. Abe, E. Watanabe, K. Sasaki, S. Iura, Surf. Coat. Technol. 198 (2005) 148. J.J. Kolodziej, T.E. Madey, Phys. Rev. B 62 (2000) 5150.
Contents lists available at ScienceDirect
Applied Catalysis B: Environmental journal homepage: www.elsevier.com/locate/apcatb
Formation of active sites on Ir/WO3–SiO2 for selective catalytic reduction of NO by CO Tetsuya Nanba *, Satoru Shinohara, Shouichi Masukawa, Junko Uchisawa, Akihiko Ohi, Akira Obuchi National Institute of Advanced Industrial Science and Technology (AIST), Research Center for New Fuels and Vehicle Technology, 16-1 Onogawa, Tsukuba 305-8569, Japan
A R T I C L E I N F O
A B S T R A C T
Article history: Received 10 January 2008 Received in revised form 18 April 2008 Accepted 22 April 2008 Available online 8 May 2008
Pretreatment conditions for the activation of Ir/WO3–SiO2 for the selective catalytic reduction of NO by CO in the presence of excess O2 were studied. Sequential treatment involving calcination in the presence of O2 and H2O followed by reduction and then re-oxidation under mild conditions was found to effectively activate Ir/WO3–SiO2. Temperature-programmed desorption during calcination, X-ray diffraction, and temperature-programmed reduction by H2 revealed that calcination was necessary for oxidative removal of the NH3 ligands from the iridium precursor, that reduction produced metallic iridium and partially reduced tungsten oxide, and that re-oxidation produced tungsten oxide with low reducibility. Transmission electron microscopy revealed that Ir was supported on finely dispersed tungsten oxide and that the iridium particle size after the sequential activation was 1–1.5 nm. ß 2008 Elsevier B.V. All rights reserved.
Keywords: Nitrogen oxide Carbon monoxide Iridium Selective catalytic reduction Tungsten oxide Silica
1. Introduction Selective catalytic reduction (SCR) of NOx (NO + NO2) by reductants present in exhaust gases is an attractive way to control NOx emissions from diesel and lean-burn engines. Since Iwamoto et al. [1] and Held et al. [2] discovered SCR by hydrocarbons (HCSCR), many papers on HC-SCR have been published [3]. Many kinds of catalysts, such as platinum-group metals [4,5], zeolites [6,7], and Al2O3-supported transition metals [8,9], exhibit HC-SCR activity. However, no HC-SCR catalytic converter system is yet in practical use. A new type of internal combustion engine is being developed for the reduction of NOx emissions: the homogeneous charge compression ignition (HCCI) engine. This engine shows low NOx and particulate emissions and relatively high-CO emissions [10]. Therefore, the use of CO as a reductant for exhausts with low-NOx concentrations is of interest. Although CO acts as a reductant in the three-way catalyst [11], it is a poor reductant under lean conditions [12,13]. Recently, iridium catalysts have been reported to show CO-SCR activity. For example, Ogura et al. reported that Ir/ silicalite at a very low-Ir loading exhibits high activity in the absence and presence of SO2 [14]. Haneda et al. reported that Ir/ SiO2 shows high activity only in the presence of SO2 [15].
* Corresponding author. Tel.: +81 29 861 8288; fax: +81 29 861 8259. E-mail address: [email protected] (T. Nanba). 0926-3373/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.apcatb.2008.04.027
Shimokawabe et al. reported that Ir/WO3 and Ir/ZnO show high activities [16] and that Ir/WO3 is effective even in the presence of SO2 [17]. Because the sulfur content in diesel fuel is now severely restricted in many countries, a catalyst having CO-SCR activity in the absence of SO2 is desirable. We have recently found that the combination of WO3 and SiO2 is an effective catalyst support [18]. The Ir/WO3–SiO2 is composed of Ir/WO3 highly dispersed on SiO2. In this study, we attempted to optimize the preparation conditions for the formation of active sites during preparation of the supported catalyst, and we discuss the active state of the Ir/WO3. 2. Experimental 2.1. Catalyst preparation The combined oxide support, WO3–SiO2, was prepared by two methods. Method 1. Addition of metallic tungsten powder (Mitsuwa Chemical Co.) to 15% H2O2 yielded a peroxopolytungstic acid solution [19]. After the pH of this solution was adjusted to approximately 8.5 with NH3, a predetermined amount of silica sol was added (Cataloid S-20L, containing 20 wt% SiO2; Catalysis & Chemical IND Co.). The mixture was dried at 110 8C and calcined in air at 500 8C for 4 h. Method 2. Ammonium tungstate parapentahydrate (Wako Pure Chemical Industries) [(NH4)10W12O41 5H2O], was dissolved in 0.7 M malic acid aqueous solution. A predetermined amount of silica sol was added, and the pH was adjusted to approximately 8.8 with NH3. Gelation occurred within
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
421
1 h after the addition of NH3. The gel was dried at 100 8C and calcined in air at 500 8C for 4 h. The WO3–SiO2 made by method 1 was prepared with WO3 contents ranging from 0% to 100%. The mixed oxide support is designated WO3(x)–SiO2-1 or -2, where the value in parentheses is the weight percentage of WO3 in the support and the next number is the preparation method. Iridium supported on WO3–SiO2 was prepared by the impregnation method using Ir(NH3)6(OH)3 as a precursor. The loading of iridium was adjusted from 0.25 to 10 wt%. The impregnated supports were calcined at 500 8C for 4 h under various atmospheres.
flow of 2% H2/Ar at a heating rate of 10 8C/min after oxidation of the sample in a 5% O2 flow at 500 8C for 2 h. The morphologies of Ir and WO3 particles were observed by transmission electron microscopy (TEM; JEM2000EX; JEOL). Temperature-programmed desorption (TPD) of nitrogen-containing compounds was carried out to clarify how the Ir precursor decomposed during these calcination procedures. After Ir loading and drying, 0.1 g of the sample was placed in the apparatus used for the activity tests. The TPD profiles were obtained from room temperature to 500 8C in a 100 mL/min flow of 10% H2, 20% O2, or 20% O2 + 1% H2O at a heating rate of 5 8C/min.
2.2. Activity tests
3. Results
Catalytic activity tests were conducted in a conventional fixedbed reactor at atmospheric pressure. The catalyst (0.1 g; sieved to between 0.15 and 0.25 mm) was diluted with quartz sand or SiC granules (particle size, 0.3–0.5 mm). The flow rate of the feed gas was 225 or 400 mL/min, corresponding to a space velocity of 34,000 or 60,000 h1, respectively. The feed gas composition for the 225 mL/min flow rate was 500 ppm NO, 5000 ppm CO, 10% O2, and 1% H2O, with He as a balance gas; for the 400 mL/min flow rate, the feed gas composition was 500 ppm NO, 3000 or 5000 ppm CO, 10% O2, and 1% or 6% H2O, with N2 as a balance gas. Before most of the catalytic activity tests, the catalysts were reduced in 10% H2 at 600 8C for 1 h. The temperature dependence of the catalytic activity was measured from 400 to 180 8C in steps of 10 8C. At each temperature, the effluent gas was sampled after the temperature had been maintained for 1 h. In the other activity test, in which measurements were carried out at a constant temperature of 270 8C, the samples were pretreated under various conditions after the H2 reduction. Concentrations of NO, NO2, N2O, CO, and CO2 in the reactant and product gases were analyzed with an FT-IR spectrometer (Magna 560; Nicolet) equipped with a multireflectance gas cell (Gemini Specialty Optics, optical path length, 2 m) and a mercury–cadmium–telluride detector. For the feed gas balanced with He, a gas chromatograph (M200; Agilent) equipped with an MS-5A PLOT column (for N2 and CO) or a PoraPLOT Q column (for CO2 and N2O) and a thermal conductivity detector was also used. The catalytic activity was evaluated in terms of NOx conversion and N2 selectivity. To ensure accuracy, N2 selectivity was evaluated only when the NOx conversion was above 10%:
3.1. Catalytic activity
NOx conversion ¼
½inlet NOx ½outlet NOx 100 ½inlet NOx
N2 selectivity ðfor He balanceÞ ¼
½outlet N2 100 ½outlet N2 þ ½outlet N2 O
N2 selectivity ðfor N2 balanceÞ ½outlet N2 O 100 ¼ 12 ½inlet NOx ½outlet NOx
2.3. Characterization The Brunauer–Emmett–Teller (BET) specific surface area was measured by N2 adsorption under a flow condition (Model 4232; Nikkiso). X-ray diffraction (XRD) patterns were measured at 40 kV and 100 mA of X-ray power (RU-300; Rigaku). Identification of crystal phase was carried out by using the JCPDS database. Temperature-programmed reduction by H2 (H2-TPR) was carried out in an adsorption test apparatus equipped with a thermal conductivity detector (ADT700; Ohkura-Riken). The TPR profiles were obtained from room temperature to 600 8C in a 100 mL/min
Table 1 lists the CO-SCR activities of 0.5 wt% Ir/WO3–SiO2-1 with various WO3 contents. At WO3 contents above 50%, the maximum NOx conversion and N2 selectivity at the temperature of maximum NOx conversion were almost the same as those of Ir/ WO3. At WO3 contents below 30%, the activity was markedly enhanced. Note that a high activity was attained even at 1 wt% WO3. WO3 contents of 1–30 wt% correspond to W/Ir molar ratios between 1.7 and 50. Note also that the precursor of WO3 had a negligible influence on the CO-SCR activity. We confirmed that the crystal phases of the WO3 in the Ir/WO3–SiO2-1 and -2 samples were the same, monoclinic and orthorhombic. The fact that the two samples were composed of the same crystal phases indicates that the WO3 precursor probably had no influence on the crystal phase. Examination of the activities at various Ir loadings showed that the maximum NOx conversion was achieved at 0.5 wt% (Table 2), and CO conversion increased with increasing Ir loading. N2 selectivity decreased with increasing Ir loading. 3.2. Effect of calcination, reduction, and additional pretreatment conditions We investigated the effects of calcination conditions on the activity of 0.5 wt% Ir/WO3(10)–SiO2-2 in terms of the maximum Table 1 Effect of WO3 content on CO-SCR activity WO3 content (wt%)
Temperature (8C)
NOx conversion (%)
N2 selectivity (%)
0 1 10 30 40 50 80 100
250 260 260 265 265 255 275 270
11 85 86 82 69 49 42 39
81 89 89 87 81 79 72 79
Catalyst: 0.5 wt% Ir/WO3(x)–SiO2-1; catalyst weight: 0.1 g; flow rate: 225 mL/min; composition: 500 ppm NO, 5000 ppm CO, 10% O2, 1% H2O, and balance He. Table 2 Effect of iridium loading on CO-SCR activity Loading (wt%)
NOx conversion (%)a
CO conversion (%)
N2 selectivity (%)
0.25 0.5 1 2 5 10
21 34 25 22 21 18
37 51 79 91 91 95
68 68 58 51 45 41
Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-2; catalyst weight: 0.1 g; flow rate: 400 mL/ min; composition: 500 ppm NO, 3000 ppm CO, 10% O2, 6% H2O, and balance N2. a Maximum NOx conversion. CO conversion and N2 selectivity were measured at 280 8C.
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
422 Table 3 Effect of calcination condition on CO-SCR activity Calcination condition
Reaction temperature (8C)
NOx conversion (%)
CO conversion (%)
N2 selectivity (%)
Air, 500 8C 20% O2, 500 8C 20% O2 + 1% H2O, 500 8C 1% O2 + 6% H2O, 500 8C 6% H2O, 500 8C 10% H2, 600 8Ca
275 260 280 275 330 290
34 7 26 31 10 4
51 81 66 72 98 97
68 – 65 68 48 –
Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-2; catalyst weight: 0.1 g; flow rate: 400 mL/min; composition: 500 ppm NO, 3000 ppm CO, 10% O2, 6% H2O, and balance N2. After calcination, sample was reduced in H2 at 600 8C for 1 h, and then activity was measured as the manner of temperature dependence. a Reduction procedure as pretreatment was omitted.
Table 4 Effect of reduction condition on CO-SCR activity Reduction condition
10% H2, Nona 10% H2, 10% H2, 10% H2,
600 8Ca b
400 8C 500 8Cb 600 8Cb
NOx conversion (%)
CO conversion (%)
N2 selectivity (%)
81 33 26 29 31
82 66 61 71 72
88 72 62 64 68
Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-1; catalyst weight: 0.1 g; calcination condition was in air at 500 8C for 4 h. Activity was measured as the manner of temperature dependence. a Flow rate: 225 mL/min; composition: 500 ppm NO, 5000 ppm CO, 10% O2, 1% H2O, and balance He. Reaction temperature was 270 8C. b Flow rate: 400 mL/min; composition: 500 ppm NO, 3000 ppm CO, 10% O2, 6% H2O, and balance N2. Reaction temperature was 280 8C.
NOx conversion as well as the CO conversion and N2 selectivity at the temperature of maximum NOx conversion (Table 3). Among the catalysts calcined in various atmospheres, the catalysts calcined in ambient air showed the highest NOx conversion. The catalyst calcined in 20% O2 showed very low activity, whereas the catalyst calcined in 20% O2 + 1% H2O showed higher activity. Furthermore, the catalyst calcined in 1% O2 + 6% H2O showed activity comparable to that of the catalyst calcined in air, whereas the catalyst calcined in 6% H2O showed low activity. The catalyst calcined in 10% H2 also showed low activity. The catalysts showing low-NOx conversions exhibited relatively high-CO conversions. These results suggest that the co-existence of O2 and H2O was necessary for activation of the catalyst. We compared the activities of reduced and unreduced catalysts after calcination and found that the reduced catalysts showed markedly higher NOx conversion than the unreduced catalysts (Table 4). The maximum NOx conversions over the catalysts reduced at 400, 500, and 600 8C were 26%, 29%, and 31%, respectively, which suggests that the reduction temperature had little effect on the activity. We investigated the effects of additional treatment conditions on the activity of 0.5 wt% Ir/WO3(10)–SiO2-2, which had been calcined in air and then reduced at 600 8C (Table 5). The catalytic activity was evaluated at 270 8C. When the catalysts were cooled
from 400 to 270 8C under a flow of CO-SCR feed gas, the catalysts showed high-NOx conversions. In contrast, cooling the catalysts under a flow of He resulted in only 7% NOx conversion. However, the catalysts subjected to additional treatment at 400 8C in 10% O2 showed approximately 50% NOx conversion. This result suggests that re-oxidation after reduction effectively activated the catalysts. However, the catalyst pretreated at 600 8C in 100% O2 was almost inactive, suggesting that mild re-oxidation was better for activation than severe re-oxidation. 3.3. Characteristics of Ir/WO3–SiO2 The specific surface area of Ir/SiO2 was 190 m2/g, and the value decreased as the WO3 content increased (Fig. 1). For the samples with 10–80 wt% WO3, the specific surface area decreased linearly with increasing WO3. The specific surface area in the range of 80– 100 wt% WO3 also linearly decreased with increasing WO3. Note that the molar amounts of WO3 and SiO2 are almost equivalent at 80 wt% WO3. The dashed line in Fig. 1 indicates the surface area of SiO2 contained in 1 g of each Ir/WO3–SiO2 catalyst. In the range from 10 to 80 wt% WO3, the slope of the curve for the SiO2 surface area is the same as the slope of the curve for the catalysts. These results suggest that the change in specific surface area in this region depended only on the SiO2 content. We measured XRD profiles of 0.5 wt% Ir/WO3(10)–SiO2-2 after calcination in air, after reduction at 600 8C, and after the CO-SCR activity test (Fig. 2). The sample calcined in air showed peaks due to WO3 (JCPDS: 43-1035, 20-1324) (Fig. 2(a)), which suggests that WO3 over SiO2 was crystalline even at 10 wt% WO3. For the sample reduced at 600 8C, the WO3 diffraction peaks disappeared, and peaks ascribed to WO3x (0 < x < 1) (JCPDS: 30-1387), WO2 (JCPDS: 32-1393), and W metal (JCPDS: 04-0806) were observed (Fig. 2(b)). For the sample subjected to CO-SCR, the XRD profile exhibited broad peaks in the WO3 region, and the WO2 and W metal peaks diminished (Fig. 2(c)). The presence of the broad peaks in the WO3 region (broader than those in Fig. 2(a)) suggests that fine WO3 particles were formed. In contrast, the sample subjected to severe oxidation conditions (100% O2 at 600 8C) showed a sharp peak at 23.18 (Fig. 2(d)), which is attributed to WO3; this treatment did not restore the peak intensity to the original level (Fig. 2(a)).
Table 5 Effect of treatment condition after reduction on CO-SCR activity Condition CO-SCR feed (introduced at 400 8C) Non CO-SCR feed (400 8C, 1 h) 10% O2 (400 8C, 1 h) 100% O2 (600 8C, 1 h)
Atmosphere before reaction CO-SCR feed He He He He
a
NOx conversion (%)
CO conversion (%)
N2 selectivity (%)
84 7 49 54 2
84 97 73 97 3
87 – 72 74 –
Activity measurement was carried out at 270 8C. Catalyst: 0.5 wt% Ir/WO3(10)–SiO2-2; catalyst weight: 0.1 g; flow rate: 225 mL/min; composition of CO-SCR feed gas: 500 ppm NO, 5000 ppm CO, 10% O2, 1% H2O, and balance He. Calcination: 500 8C for 4 h in air; reduction: 600 8C for 1 h in 10% H2. a Temperature dependence procedure as shown in Fig. 1(a).
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
423
Fig. 1. Dependence of BET surface area on WO3 content for 0.5 wt% Ir/WO3(x)– SiO2-1.
Although crystallization of WO3 seemed to be promoted by the severe oxidation conditions, the WO3 particles obtained after calcination, reduction, and severe re-oxidation must have been smaller than those obtained after calcination alone. We also examined the H2-TPR profiles for various 0.5 wt% Ir catalysts (Fig. 3). Ir/SiO2 exhibited a reduction peak at 225 8C, which was assigned to the reduction of IrO2 to Ir metal. Ir/WO3 had a broad peak centered at 440 8C, in addition to a peak at 220 8C, whereas WO3 had no peak below 550 8C. The reduction temperature of WO3 is known to be lowered in the presence of a noble metal [20,21]. Therefore, we ascribed the broad peak at around 440 8C to the reduction of WO3. H2 consumption values for the peaks at around 220 8C for Ir/SiO2 and Ir/WO3 were 42 and 310 mmol/g, respectively. These values correspond, respectively, to 0.8 and 6 times the calculated H2 consumption for
Fig. 2. XRD profiles of 0.5 wt% Ir/WO3(10)–SiO2-2 treated under various conditions: (a) after calcination in air, (b) after reduction at 600 8C, (c) after re-oxidation under CO-SCR conditions, and (d) after re-oxidation at 600 8C in 100% O2.
Fig. 3. H2-TPR profiles of various 0.5 wt% Ir catalysts.
stoichiometric reduction of IrO2 to Ir. The high-H2 consumption (substantially exceeding unity) suggests that IrO2 reduction over WO3 proceeded together with reduction of WO3. Ir/WO3(10)– SiO2-2 exhibited H2 reduction peaks at 225 and 350 8C. The amount of H2 consumed below 500 8C corresponded to the formation of WO2.9. Additional H2 was consumed above 550 8C, which suggests that tungsten oxides were reduced further at high temperatures. Re-oxidized Ir/WO3(10)–SiO2-2 exhibited a H2 reduction peak at 190 8C, which we ascribed to the reduction of IrO2 and WO3; no other peak appeared below 550 8C, which suggests that the reduction of tungsten oxides was suppressed after the re-oxidization. We investigated the H2-TPR profiles of Ir/WO3–SiO2 catalysts with various WO3 contents (Fig. 4(a)). At WO3 contents below 10 wt%, the TPR profiles showed a peak at 235 8C with a shoulder at 200 8C. At WO3 contents above 10 wt%, the shoulder at 200 8C increased and shifted toward lower temperatures as the WO3 content increased. We determined the molar ratio of H2 consumed to Ir atoms for the peak at around 220 8C, including the shoulder (or peak) at around 200 8C, and examined the correlation between this ratio and the WO3 content in 0.5 wt% Ir/WO3–SiO2-1 after calcination (Fig. 4(b)). A ratio of 2 suggests stoichiometric reduction of IrO2 to Ir. Below 30 wt% WO3, the ratio was slightly lower than 2 at 30 wt% WO3, the ratio was slightly higher than 2, and as the WO3 content increased, the ratio increased markedly and substantially exceeded 2. This observed behavior for the peak and the shoulder at > 30 wt% WO3 indicates that the reducibility of WO3 in the vicinity of IrO2 was higher for catalysts containing >30 wt% WO3 than for catalysts containing less WO3. Lucas et al. reported that the reducibility of WO3 decreases with decreasing WO3 content [22]. These results suggest that reduction of WO3 in the vicinity of IrO2 did not occur to a substantial extent for catalysts with WO3 contents below 30 wt% and that the reduction of WO3 increased with increasing WO3 content at values above 30 wt%.
424
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
Fig. 4. H2-TPR profiles of the Ir/WO3–SiO2-1 (a) and the molar ratio of H2 consumed to Ir atoms for the peak at 220 8C and the dependence of the ratio on WO3 content (b).
Fig. 5. TEM image of 0.5 wt% Ir/WO3(10)–SiO2-2 treated by optimal sequential treatment (a) and reduced without calcination (b).
TEM images of 0.5 wt% Ir/WO3(10)–SiO2-2 subjected to CO-SCR conditions showed that cubic tungsten oxides with particle sizes larger than 5 nm were dispersed on SiO2 and that Ir particles 1– 1.5 nm in size were located on the tungsten oxides (Fig. 5(a)). In contrast, TEM images of the reduced 0.5 wt% Ir/WO3(10)–SiO2-2 without calcination, which had low activity, exhibited WO3 particles of various sizes and Ir particles 2–5 nm in size (Fig. 5(b)). Furthermore, there were some Ir particles that were not associated with WO3 particles, as indicated by the absence of a lattice pattern of WO3 around the Ir particles. We concluded that the final active form of Ir/WO3–SiO2 obtained by sequential treatment involving calcination, reduction, and re-oxidation was very fine Ir/WO3 particles dispersed on SiO2. 4. Discussion The preferred composition of Ir/WO3–SiO2 for CO-SCR was 1– 30 wt% WO3 in the support and a 0.5 wt% Ir loading. The surface area of Ir/WO3–SiO2 was linearly correlated with SiO2 content. In contrast, almost equivalent activities of the samples with 1– 30 wt% WO3 loading were observed. Previously, we reported that iridium particles existed selectively on the WO3 particles, and we supposed that the increase in the surface area of Ir/WO3 by dispersion on the SiO2 resulted in the enhancement of the CO-SCR activity [18]. However, these results suggest that the enhancement of the CO-SCR activity of Ir/WO3–SiO2 at lower WO3 contents was due not to an increase in the Ir/WO3 surface area but to some other reason, which is discussed below.
Fig. 6. Temperature-programmed desorption of nitrogen-containing compounds during calcination under 10% H2, 20% O2, and 20% O2 + 1% H2O. Catalyst weight, 0.1 g; flow rate, 400 mL/min; ramping rate, 5 8C/min.
T. Nanba et al. / Applied Catalysis B: Environmental 84 (2008) 420–425
The change in CO-SCR activity with WO3 content agreed well with the change in the reduction properties of IrO2 and WO3 at 220 8C as revealed by H2-TPR (Fig. 4). This result suggests that what makes WO3 preferable as a support for Ir is its resistance to reduction when in the vicinity of Ir particles. To activate Ir/WO3–SiO2, the catalysts had to be treated sequentially by calcination (oxidation in the presence of H2O), reduction, and re-oxidation. TEM observations indicated that the reduced sample without calcination had larger Ir particles than the sample subjected to the optimal sequential treatment and that the former sample had separate Ir and WO3 particles. Ir/SiO2 is known to exhibit activity only in the presence of SO2 [15], so the decrease in the CO-SCR activity of the reduced catalysts without calcination was probably due to dissociation of Ir from WO3. To observe the decomposition behavior of Ir precursor, we used TPD to measure the formation of nitrogen-containing compounds from WO3–SiO2 impregnated with Ir(NH3)6(OH)3 (Fig. 6). NOx and N2O formed at 375 8C in 20% O2. Furthermore, the addition of H2O to 20% O2/N2 led to the evolution of NOx at a lower temperature of approximately 335 8C. This result suggests that calcination in the presence of O2 and H2O resulted in the H2O-promoted oxidative removal of ligands of Ir(NH3)6(OH)3. On the other hand, calcination under H2/ N2 resulted in NH3 formation between 200 and 300 8C, which suggests the simple release of NH3 ligands. Considering that H2 reduction after calcination was performed in the absence of NH3 ligands for the active Ir/WO3–SiO2, the residual NH3 ligands might relate to the dissociation of Ir form WO3 in the reducing conditions. The role of calcination is suspected to be the oxidative removal of NH3 ligands. H2 reduction after calcination was necessary for activating Ir/ WO3–SiO2. Metallic iridium is believed to be the active site for HCSCR [23,24] and CO-SCR [15], and the reduction procedure is necessary for the formation of metallic iridium. Moreover, the reduction procedure markedly changed the oxidation state of tungsten, as shown by the XRD results. Although WO3 is known to form alloys with various metals upon reduction [21] and an Ir–W alloy has been shown to have high resistance against oxidation [25], 600 8C is too low a temperature for the formation of an Ir–W alloy [26]. We observed only a small difference between the activities of the samples reduced at 400 8C to 600 8C, and we also confirmed that W metal and WO2 were not observed by XRD after reduction at 400 8C. This result suggests that severe reduction to form W and WO2 was not dispensable. H2 consumption below 500 8C in H2-TPR of Ir/WO3(10)–SiO2-2 was insufficient for the formation of WO2. Therefore, we suggest that the necessary oxidation state of tungsten after reduction is WO3x (0 < x < 1). We believe that the re-oxidation step was necessary for the formation of WO3 with low reducibility as a support for iridium. Although the H2-TPR profile of Ir/WO3–SiO2 after calcination showed a broad peak at 350 8C, the peak disappeared in the H2TPR profile of Ir/WO3–SiO2 after re-oxidation. H2 consumption at around 200 8C in the H2-TPR measurement of the re-oxidized sample was about 0.4 times the amount expected for stoichiometric IrO2 reduction. These results suggest that the iridium particles were composed mainly of metallic iridium and that the reducibility of WO3 was decreased by re-oxidation. Bigey et al. have reported that re-oxidized tungsten oxides exhibit lower reducibility than fresh WO3 [21]. It is known that the oxidation state of tungsten is easily changed during reactions such as cracking and isomerization of hydrocarbons and that retaining a stable oxidation state of tungsten with high activity is difficult [20]. We suspect that the decrease in the reducibility of WO3 was important for maintaining an active metallic iridium surface. It is concluded that the final active site formed by sequential treatments was metallic-state iridium finely dis-
425
persed on WO3 having small particle size and resistance to reduction. 5. Conclusions Ir/WO3–SiO2 showed high activity for selective NO reduction by CO. We attempted to optimize the conditions for preparation of the catalyst. The optimum composition was 0.5 wt% Ir and 1–30 wt% WO3 in the support. At the optimum WO3 contents, NOx conversions were almost the same for all WO3 contents. The surface area of Ir/WO3–SiO2 depended on SiO2 content. Although the combination of WO3 and SiO2 as a support was effective for dispersing Ir/WO3, the surface area did not influence the CO-SCR activity. The H2-TPR results indicated that the catalysts with the optimum WO3 content had lower reducibility of WO3 than the catalysts with a high-WO3 content. Pretreatment after Ir loading (that is, calcination, reduction, and additional treatment) strongly influenced the CO-SCR activity. The optimum sequence of pretreatment was calcination in the presence of O2 and H2O followed by H2 reduction and then reoxidation under mild oxidizing conditions. TPD of nitrogencontaining compounds during calcination suggested that the NH3 ligands in the Ir precursor were removed by oxidation in the presence of O2 but that NH3 release under a reducing condition occurred at a lower temperature than oxidative removal of the NH3 ligands. H2O addition lowered the temperature for oxidative removal of the NH3 ligands. During the reduction, iridium was reduced to the metallic state, and WO3 was also reduced to metal and WO2. The reduction of WO3 at least to WO3x (0 < x < 1) was required to activate the catalyst. H2-TPR results suggested that reoxidation suppressed the reducibility of WO3. The re-oxidation step formed the favorable state of tungsten oxide as a support for iridium. XRD and TEM results suggested that very small WO3 particles were dispersed on SiO2. We concluded that the active state of the Ir/WO3–SiO2 catalysts, which was formed by sequential treatments, was metallic-state iridium finely dispersed on WO3, which had small particle size and resistance to reduction. References [1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25] [26]
M. Iwamoto, H. Yahiro, S. Shundo, Y. Yu-u, N. Mizuno, Appl. Catal. 69 (1991) L15. W. Held, A. Ko¨nig, T. Richter, L. Puppe, SAE Tech. Pap. Ser. No. 900496 (1990). V.I. Paˆrvulescu, P. Grange, B. Delmon, Catal. Today 46 (1998) 233. A. Obuchi, A. Ohi, M. Nakamura, A. Ogata, K. Mizuno, H. Ohuchi, Appl. Catal. B 2 (1993) 71. R. Burch, T.C. Watling, J. Catal. 169 (1997) 45. J.L. d’Itri, W.M.H. Sachtler, Catal. Lett. 15 (1992) 289. S. Sato, Y. Yu-u, H. Yahiro, N. Mizuno, M. Iwamoto, Appl. Catal. 70 (1991) L1. H. Hamada, Y. Kintaichi, M. Inaba, M. Tabata, T. Yoshinari, H. Tsuchida, Catal. Today 29 (1996) 53. Y. Torikai, H. Yahiro, N. Mizuno, M. Iwamoto, Catal. Lett. 9 (1991) 91. D. Jun, K. Ishi, N. Iida, JSME Int. J. Ser. B 46 (2003) 60. J.T. Kummer, J. Phys. Chem. 90 (1986) 4747. M.L. Unland, J. Phys. Chem. 77 (1973) 1952. M.L. Unland, J. Catal. 31 (1973) 459. M. Ogura, A. Kawamura, M. Matsukata, E. Kikuchi, Chem. Lett. (2000) 146. M. Haneda, Pursparatu, Y. Kintaichi, I. Nakamura, M. Sasaki, T. Fujitani, H. Hamada, J. Catal. 229 (2005) 197. M. Shimokawabe, N. Umeda, Chem. Lett. 33 (2004) 534. M. Shimokawabe, M. Niitsu, H. Inomata, N. Iwasa, M. Arai, Chem. Lett. 34 (2005) 1426. T. Nanba, S. Shinohara, J. Uchisawa, S. Masukawa, A. Ohi, A. Obuchi, Chem. Lett. 35 (2006) 450. J. Oi, A. Kishimoto, T. Kudo, J. Solid State Chem. 103 (1993) 176. C. Bigey, L. Hilaire, G. Maire, J. Catal. 198 (2001) 208. C. Bigey, L. Hilaire, G. Maire, J. Catal. 184 (1999) 406. A. de Lucas, J.L. Valverde, P. CanA¨izares, L. Rodriguez, Appl. Catal. A 172 (1998) 165. M. Nawdali, E. Iojoiu, P. Ge´lin, H. Praliaud, M. Primet, Appl. Catal. A 220 (2001) 129. C. Wo¨gerbauer, M. Maciejewski, A. Baiker, J. Catal. 205 (2002) 157. Y. Abe, E. Watanabe, K. Sasaki, S. Iura, Surf. Coat. Technol. 198 (2005) 148. J.J. Kolodziej, T.E. Madey, Phys. Rev. B 62 (2000) 5150.